Cisco-Platform Settings

System and Other Tuning Information

Platform settings

One or more of the following settings may have been set. If so, the "General Notes" section of the report will say so; and you can read below to find out more about what these settings mean.

Linux Huge Page settings

In order to take advantage of large pages, your system must be configured to use large pages. To configure your system for huge pages perform the following steps:

Note that further information about huge pages may be found in your Linux documentation file: /usr/src/linux/Documentation/vm/hugetlbpage.txt

Transparent Huge Pages

On RedHat EL 6 and later, Transparent Hugepages increase the memory page size from 4 kilobytes to 2 megabytes. Transparent Hugepages provide significant performance advantages on systems with highly contended resources and large memory workloads. If memory utilization is too high or memory is badly fragmented which prevents hugepages being allocated, the kernel will assign smaller 4k pages instead. Hugepages are used by default if /sys/kernel/mm/redhat_transparent_hugepage/enabled is set to always

HUGETLB_MORECORE

Set this environment variable to "yes" to enable applications to use large pages.

LD_PRELOAD=/usr/lib64/libhugetlbfs.so

Setting this environment variable is necessary to enable applications to use large pages.

KMP_STACKSIZE

Specify stack size to be allocated for each thread.

KMP_AFFINITY

KMP_AFFINITY = < physical | logical >, starting-core-id
specifies the static mapping of user threads to physical cores. For example, if you have a system configured with 8 cores, OMP_NUM_THREADS=8 and KMP_AFFINITY=physical,0 then thread 0 will mapped to core 0, thread 1 will be mapped to core 1, and so on in a round-robin fashion.

KMP_AFFINITY = granularity=fine,scatter
The value for the environment variable KMP_AFFINITY affects how the threads from an auto-parallelized program are scheduled across processors.
Specifying granularity=fine selects the finest granularity level, causes each OpenMP thread to be bound to a single thread context.
This ensures that there is only one thread per core on cores supporting HyperThreading Technology
Specifying scatter distributes the threads as evenly as possible across the entire system.
Hence a combination of these two options, will spread the threads evenly across sockets, with one thread per physical core.

OMP_NUM_THREADS

Sets the maximum number of threads to use for OpenMP* parallel regions if no other value is specified in the application. This environment variable applies to both -openmp and -parallel (Linux and Mac OS X) or /Qopenmp and /Qparallel (Windows). Example syntax on a Linux system with 8 cores: export OMP_NUM_THREADS=8

Hardware Prefetch:

This BIOS option allows the enabling/disabling of a processor mechanism to prefetch data into the cache according to a pattern-recognition algorithm.

In some cases, setting this option to Disabled may improve performance. Users should only disable this option after performing application benchmarking to verify improved performance in their environment.

Adjacent Sector Prefetch:

This BIOS option allows the enabling/disabling of a processor mechanism to fetch the adjacent cache line within an 128-byte sector that contains the data needed due to a cache line miss.

In some cases, setting this option to Disabled may improve performance. Users should only disable this option after performing application benchmarking to verify improved performance in their environment.

Intel Turbo boost Technology:

Enabling this option allows the processor cores to automatically increase its frequency and increasing performance if it is running below power, temperature.

Intel Hyper Threading Technology:

Enabling this option allows to use processor resources more efficiently, enabling multiple threads to run on each core and increases processor throughput, improving overall performance on threaded software.

Data Reuse Optimization:

Enabling this option allows very active cache lines to stay in the L2 by sending a hint to the L3 to prevent them from aging even if they are not heavily used in the L3.

Depending on the specific bios version being used, this option may be called "LRU Hint".

High Bandwidth:

Enabling this option allows the chipset to defer memory transactions and process them out of order for optimal performance.

ulimit -s <n>

Sets the stack size to n kbytes, or unlimited to allow the stack size to grow without limit.

numactl --interleave=all "runspec command"

Launching a process with numactl --interleave=all sets the memory interleave policy so that memory will be allocated using round robin on nodes. When memory cannot be allocated on the current interleave target fall back to other nodes.

Free the file system page cache

The command "echo 1> /proc/sys/vm/drop_caches" is used to free up the filesystem page cache.

submit= MYMASK=`printf '0x%x' $((1<<$SPECCOPYNUM))`; /usr/bin/taskset $MYMASK $command

When running multiple copies of benchmarks, the SPEC config file feature submit is sometimes used to cause individual jobs to be bound to specific processors. This specific submit command is used for Linux. The description of the elements of the command are:

Using numactl to bind processes and memory to cores

For multi-copy runs or single copy runs on systems with multiple sockets, it is advantageous to bind a process to a particular core. Otherwise, the OS may arbitrarily move your process from one core to another. This can effect performance. To help, SPEC allows the use of a "submit" command where users can specify a utility to use to bind processes. We have found the utility 'numactl' to be the best choice.

numactl runs processes with a specific NUMA scheduling or memory placement policy. The policy is set for a command and inherited by all of its children. The numactl flag "--physcpubind" specifies which core(s) to bind the process. "-l" instructs numactl to keep a process memory on the local node while "-m" specifies which node(s) to place a process memory. For full details on using numactl, please refer to your Linux documentation, 'man numactl'

submit= $[top]/mysubmit.pl $SPECCOPYNUM "$command"

On Xeon 74xx series processors, some benchmarks at peak will run n/2 copies on a system with n logical processors. The mysubmit.pl script assigns each copy in such a way that no two copies will share an L2 cache, for optimal performance. The script looks in /proc/cpuinfo to come up with the list of cores that will satisfy this requirement. The source code is shown below.

Source
******************************************************************************************************


#!/usr/bin/perl
 
use strict;
use Cwd;
 
# The order in which we want copies to be bound to cores
# Copies: 0, 1, 2, 3
# Cores:  0, 1, 3, 6
 
my $rundir        = getcwd;
 
my $copynum = shift @ARGV;

my $i;
my $j;
my $tag;
my $num;
my $core;
my $numofcores; 

my @proc;
my @cores;

open(INPUT, "/proc/cpuinfo") or
   die "can't open /proc/cpuinfo\n"; 

#open(OUTPUT, "STDOUT");

# proc[i][0] = logical processor ID
# proc[i][1] = physical processor ID
# proc[i][2] = core ID

$i = 0;
$numofcores = 0;

while(<INPUT>)
{
  chop;
 
  ($tag, $num) = split(/\s+:\s+/, $_);


  if ($tag eq "processor") {
      $proc[$i][0] = $num;
  }

  if ($tag eq "physical id") {
      $proc[$i][1] = $num;
  }

  if ($tag eq "core id") {
      $proc[$i][2] = $num;
      $i++;
      $numofcores++;
  }
}

$i = 0;
$j = 0;

for $core (0, 4, 2, 1, 5, 3) {
  while ($i < $numofcores) {
     if ($proc[$i][2] == $core) {
        $cores[$j] = $proc[$i][0];
        $j++;
     }
     $i++;
  }
  $i=0;
}

open  RUNCOMMAND, "> runcommand" or die "failed to create run file";
print RUNCOMMAND "cd $rundir\n";
print RUNCOMMAND "@ARGV\n";
close RUNCOMMAND;
system 'taskset', '-c', $cores[$copynum], 'sh', "$rundir/runcommand";